DFlow

You can access Kalshi-style prediction market data through this MCP server, which exposes a comprehensive set of endpoints for events, markets, trades, forecasts, candlesticks, and live data. It’s designed to let MCP clients retrieve metadata, market information, and analytics in a standardized way, enabling seamless integration with your trading and forecasting workflows.

How to use

To use this MCP server with an MCP client, you can connect either via a remote HTTP endpoint or by running a local stdio-based server. The remote HTTP URL provides a centralized API you point your client at, while the local stdio option runs the server on your machine and communicates over standard input/output. Your client can query events, markets, trades, forecast histories, candlesticks, and live data using the provided tool interfaces. If you installed a local instance, start it in production mode and configure your client to connect via the stdio transport (command and arguments you supply determine how the client starts the process). For Claude Desktop or other MCP clients, provide the server configuration so the client knows how to start or reach the MCP server. Remember to supply the correct API end-point or process command when configuring your client.

How to install

You have two installation paths depending on your preference for a hosted or a local setup.

Step-by-step setup for HTTP access

Use the HTTP endpoint to access the MCP API directly from your MCP client. This provides a centralized, remote API surface for all supported tools and data streams.

Step-by-step setup for stdio access

Run a local MCP server instance and connect via the stdio transport. This approach lets you run the server on your machine and have clients communicate through standard input/output channels.

Interacting with common data endpoints

• Retrieve events and event metadata to understand the available forecasting windows and markets.

Security considerations

If you expose the HTTP endpoint publicly, ensure appropriate access controls and consider API authentication mechanisms. For local stdio usage, keep the process secured within your environment and implement least-privilege file and network access.

Available tools

get_event

Fetch a single event by ticker, with optional nested market data for context.

get_events

Retrieve a paginated list of events, with support for filtering and sorting.

get_market

Obtain detailed information about a market by its ticker.

get_market_by_mint

Look up a market by its mint address to retrieve related data.

get_markets

List markets with pagination and filtering options.

get_markets_batch

Fetch multiple markets in a single request (up to 100).

get_trades

Get trades across markets, with optional time-window filtering.

get_trades_by_mint

Retrieve trades for a specific market identified by its mint.

get_forecast_percentile_history

Obtain forecast percentile history for a given series and event.

get_forecast_percentile_history_by_mint

Forecast history by mint for a specific series.

get_event_candlesticks

Retrieve OHLC candlestick data for an event.

get_market_candlesticks

Retrieve OHLC candlestick data for a market.

get_market_candlesticks_by_mint

Candlesticks data for a market identified by mint.

get_live_data

Fetch live milestone data for ongoing events.

get_live_data_by_event

Live data stream for a specific event.

get_live_data_by_mint

Live data stream for a market identified by mint.

get_series

List all series templates available for forecast and analytics.

get_series_by_ticker

Retrieve a specific series by its ticker.

get_outcome_mints

Return outcome mint addresses for a given event or market.

filter_outcome_mints

Filter the set of mints to only include outcome mints.

get_tags_by_categories

Provide a mapping of categories to tags for filtering.

get_filters_by_sports

Return available filtering options by sport or category.

search_events

Search events by title or ticker with sorting and limits.